how does a tester add value in this new construct? · automation •we cant do devops without...

Post on 04-Aug-2020

0 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

How does a tester add value in this new construct?• A Call to Action!

• Understand the stack

• Understand the tools

• Understand the types of testing

• Understand the customer

• Understand where to contribute

• Irrespective of how your manager or team works, go forth!

Automation

• We can’t do DevOps without Automation (test, and other types):

• Mike D. Kail: “Automation is a key component”

• John Arundel: “Without automation, there is no DevOps”

• Gregory S. Bledsoe: “Collaboration and automation are the two foundational elements of DevOps. Without automation, what you are doing isn’t DevOps”

• Thorsten Heller: “DevOps requires automation”

• Eric Vanderburg: “Automation is critical in DevOps”

• Quentin Adam: “ It’s one of our core values ”...

Test Automation

Good Tests Are….

• Implemented the “right way”:• If they’re GUI based: Tolerant to changes• Modularized• Data Driven• De-buggable• Version Controlled• Stored with the code they’re testing• De-coupled from environment• Use standard libraries• Managed data as part of the test (sets itself up)• Value of the test is continually reviewed• …

Technical Debt

• In our struggle to balance:• Build the right thing

• Build the thing right

• Build it in time

• Typically, we build tests in the wrong place

Technical Debt

• Tests are a derivative artifact

• Not dissimilar to a 'technical debt' - every change made requires more work to maintain tests

• Create items in your backlog to demonstrate the cost and benefit of paying it off

• "Push down the pyramid"

Why pay the debt?

• Moving the tests closer to where the function was developed

• To exercise the function more quickly

• To receive better feedback on change

• Lower maintenance costs

Manual Testing (top of the pyramid)

• Manual testing is NOT a bad thing

• “A repeatable test strategy can minimize the chance that you’ll discover new problems. A variable test strategy will at least not avoid finding bugs”• James Bach

• The same argument can be made about Automated Tests

Manual Testing

• Actually, there's a lot of manual testing that goes on

• Exploratory, Business Representation, Walkthroughs/Showcases (demos are great)

• Our job is to make this manual testing AS EFFICIENT AS POSSIBLE!• Only manually test what needs to be manually tested

Where do we put it?

• With EVERYTHING else - in version control• Code

• Configuration

• DB Schemas

• Tests

• Infrastructure

Version Control (probably GIT)

• There’s going to be:• Branches• Merges• Pull Requests (aka merge requests)• Triggers• Tags• Squashing• Rebasing• Cloning• Forking• …

• It’s going to be GREAT!

This talk isn’t about Git

• However, why so important?• Single source of truth

• Everyone knows where to look

• Pull/Merge requests – Governance around peer reviews

• Testers get a view of code, and developers get a view of tests

• Full history of everything

Not just ‘Test’ Automation

• Infrastructure Automation and Verification (testing)• Provisioning

• Configuring

• Orchestrating

• Checking

It’s all about the STACK

Infrastructure as Code (IaC) is cool ☺

• Docker• Kubernetes• OpenShift• Puppet• Chef• Terraform• Ansible• SaltStack• Service Virtulisation• …

Capture the recipe

• We address the problem of "Environmental drift”

• Environments are available – no downtime

• Aim for production to be same as every other environment

• Enables parallel testing (tests don’t pollute each other’s environment)

• We approach a situation where we have true, ephemeral environments

• OPS are here to help!

Test the Infrastructure

• How do we know that our ephemeral environment is ‘orchestrated’ properly?

• Write a test that proves the infrastructure is good – against the recipe• https://serverspec.org/

• Execute your test as part of the pipeline

The Pipeline…

• Continuous Integration: • Everyone commits every day• Confident (and demonstrable) coverage of tests

• Continuous Delivery: • We *can* go to production, but we choose not to, the business might want intervene

• Continuous Deployment:• Every change goes to production

A “Pipeline”

PIPELINE – 10 minutes long

Test itBuild it, and maybe put it somewhere

Monitor it

Put it somewhere

The Pipeline…

• 10 Minute Build:

• “The 10-minute build practice is an extreme programming practice where the code base is designed by the developer to be built automatically. The code base is also designed to test run in ten minutes or less. It is from the amount of time required for the code base to finish running all tests that the 10-minute build derives its name.”

https://explainagile.com/agile/xp-extreme-programming/practices/10-minute-build/

Tools: A shared and interoperable ecosystem

• Wiki

• Ticket system (force ranked backlog)

• Source Control

• CI Pipeline

• Binary Storage

• Infrastructure

Methods…

• Iterate

• Definition of Done

• Definition of Ready

• Single (awesome) Product Owner

• Visibility

• Communication

• Measurement

Measurement

• Metrics are VITAL for improving things – we do more with DevOps

• Build them into your process

• Monitor production too

• Learn and change, but not on a whim

How does this work?

Wiki

Version

ControlTickets CI Pipeline

Binary Store

Wiki

Wiki:

• English words

• Collaboration

• Versioning

• Transparency

• Templates

Epic

Epic

Feature#1 Feature#2

Meetings

Retro

Tickets

Tickets:

• Scrum/Kanban/etc…

• Captures data

• Visible/Radiator

• Definition of Done

• Integrated

• Kicks off feature branches

Wiki

TODO

#4

#5

DOING

#3

DONE

#2

#1

Wiki

Version

ControlTickets

Version Control:

• Ticket == Branch

• Branching Strategy

• Every commit diff back to Ticket

Wiki

Version

ControlTickets

Pipeline:

• Triggered on every commit

• …to every branch

• Feed back results of the pipe

• Guard the DoD

CI Pipeline

CI Pipeline

CI Pipeline

And finally…

Questions?

top related